Background and objective: There is a paucity of survival data regarding the prognosis of elderly patients following acute exacerbations of COPD (AECOPD). We undertook a study to examine long-term mortality rates and to identify clinical and laboratory predictors of these outcomes. Methods: A retrospective cohort study was conducted of 786 consecutive elderly (>65 years) patients admitted to general medicine acute-care wards for AECOPD. Factors determining short- and long-term mortality were analysed. Results: The mean (±SD) age of the study population was 75.8 ± 7.3 years (range 65-100 years). The in-hospital mortality rate for the entire cohort was 7.25%. The risk of mortality at 1, 3 and 5 years was 28%, 47% and 54%, respectively. In univariate analysis age (hazard ratio 1.52; 95% confidence interval: 1.23-1.91), FEV 1 (1.45; 1.73-2.35), active cancer (1.23; 1.64-2.32), current smoking (1.74; 1.35-2.11), ischaemic heart disease (1.58; 1.28-2.02), congestive heart failure (1.55; 1.23-2.26) and maintenance use of oral glucocorticosteroids (1.58; 1.11-2.79) were significantly associated with mortality. In multivariate analysis, only current smoking (1.89; 1.18-1.93), ischaemic heart disease (1.41; 1.07-1.68), PaCO2 on admission (1.49; 1.03-1.60), hospital readmission (2.23; 1.40-2.18) and FEV1 (1.41; 1.12-1.54) were independent predictors of mortality. Conclusions: This study provides new insights into the predictive factors associated with long-term prognosis in elderly patients admitted for acute exacerbations of COPD, which differ from those previously identified for younger patients.